What is evidence?

Popper tells us we should judge theories by how well they predict or explain.

He also says we can reject theories if they predict the wrong thing.

Some scientists, including a good many palaeontologists, say that's no good to them, they need proof of something before they can accept it, and at the very least, good evidence. They can't be messing with just predictions and explanations. Without real evidence, they say, it's all just metaphysics.

Happily, it turns out "evidence" is composed of nothing but prediction and explanation, just as atoms are composed of sub-atomic particles...

An observation is evidence for a theory if they are compatible with each other, ie the theory predicts and possibly explains the observation:

But compatibility isn't necessarily very strong evidence if other theories are just as compatible.

However, if competing theories don't predict the observation, or predict a different one, then the observation is stronger evidence for the first theory:

To the extent that it can be demonstrated that ALL possible alternative theories predict something other than the observation, the observation will be proof of the first theory. Of course the demonstration required is seldom possible.


Hypotheses as candidate solutions, and their use in science.

Even if the world is real, and I assume that it is, science involves both reality and the mind.

In the mind we have a number of things:

eg basic sensations:

[blue], [roundness], [a type of smooth motion (one of many)].

Some things are more complex, being made up of basic sensations:

Much of our knowledge is in the form of compound concepts such as this.

To save time and avoid reinventing the wheel every time we see something familiar, we learn how to check just enough of what we see to know which compound concept we're talking about.

Thinking, and science, which perhaps we may consider as a type of thinking, largely involve the way we use our compound concepts.

The compound concept above was used in perception - perceiving a wheel.

A compound concept can also be used as a plan. This would come from a sequence of actions that have proved useful in the past in some circumstance.

Althogh compound concepts when used in perception basically work from bottom-up, they can also work from top-down. Thus, if we decide to step off the street for a MacDonalds, we develop a compound idea into its sequence of component actions.

Of course, in a complex being, the plans are very flexible, have many layers, and some bits involve embedded perceptions. They will often be fairly new, and be weakly structured compared to other groupings of elementary and compound concepts, so the whole mental operation will be like a slime mould, elements grouping together into one compound concept here, then splitting and behaving as if part of another one, there.


Here's how, in my view, science is involved:

When we try to find out something, say, how a particular object moves about, we start with a small amount of knowledge, and a number of as yet unconfirmed possibilities come into our minds in the form of high-level compound perception concepts. They won't at this stage have all their little downward branches attached to anything real though, and the high-level concepts themselves will therefore be unreal too. Thus, if we had an idea that woodlice might move about on wheels, we would have the "wheels" concept in our minds, but in unreal mode. We would then try to check out each of the downward branches to see if they were real. When we had checked the reality of enough sub-concepts, we would decide woodlice either did, or didn't have wheels.

Note here, people have a strong tendency to confirm hunches as true. They have to otherwise we would get stuck and make no decisions. It is really a form of "we will TREAT it as true", rather than believing it is absolutely true. Within thought, deciding what to do is much more important than deciding what is true. Thoughts are the servant of actions in beings that survive well. Actions can be definitely made or definitely not made; theories on the other hand can be only "disproved" or "not yet disproved".

Having flipped our woodlouse upsidedown, we decide that having looked this carefully for this long and not having found any elementary features of wheels, means we are happy they don't have them. However, our consciousness is now flooded with perceptual subcomponents of the "legs" compound concept. Working in a bottom-up manner, they force the legs compound concept into our minds in a state confirmed as real.

These compound concepts are the cause of a lot of misunderstanding sciencewise. They are essential for advanced thought - it has been shown that the best plan for perception involves making guesses and then seeing if they are right or not. Any alternative is far too slow if it could be made to work at all. (Often of course what's in front of us is new, so we have to invent a new concept.) However, these compound concepts often come to mind before we have enough evidence to say whether they are true - they have to otherwise we wouldn't be using the "generate a guess and test it" strategy well. Also, we would not survive well; a loud rumbling from round the corner may not be confirmed as an elephant or bus but we must be in the habit of entertaining the possibility that it is.

Compound concepts merely reflect patterns of other concepts that have occured frequently or significantly enough in the past. They may be highly abstract; they might only reflect certain aspects of something more concrete. These aspects may then be used with profit as an analogy in other areas. Use of analogy is another essential to advanced thought. The analogy of a plug-hole occured to someone once, and he used it to help formulate the concept of the black hole. The *behaviour* of water going down a plug-hole appeared in someone's mind in an unreal state, and he checked to see if it applied in astrophysics. Evidence for light behaving in that way was not available in full detail before it was checked. However there is no point in saying before one has checked the evidence that one doesn't believe in black holes.

Claiming that as a positivist one doesn't believe in something for which one has no firm evidence is easy to say for someone who hasn't tried to develop a system showing significant thought. Restricting oneself to "I believe" or "I don't believe" isn't enough; one also needs "I suspect" along with a lot of other subtleties, if one is either creating a thinking machine, or partaking in science to as full an extent as one might. Hawking has mastered the skills to an excellent level subconsciously, but that doesn't let other people (or even himself) off understanding them consciously.

On the question of the worthiness of logical positivism, it was the stock in trade of the Vienna circle who were inspired by the early Wittgenstein. He changed his mind a good deal over the years; it might be interesting to investigate whether he died a logical positivist. AJ Ayer joined the Vienna circle briefly and became a logical positivist, but he definitely changed his mind and recognised he was wrong. Popper of course was one of their main antagonists; his comments in Cs & Rs show hilariously how he thinks he only confused them (as mentioned in my website) but statistical operations of importance (ie where money is involved) use the rejection of hypotheses, not their confirmation, as the basic principle - ie Popperism embodied.


"Metaphysics" as a disparaging term, and modern knowledge science.

My comments on the philosophy of science particularly as regards the use of unconfirmed hypotheses are often greeted implicitly or otherwise with disparaging use of the word "metaphysics". To those who might be in this category I would recommend the website: http://pages.cpsc.ucalgary.ca/~kremer/courses/CG/CGlecture_notes.html
up to and including about section 3.2. Here you will find my concepts and methods repeated, with references. Note how old some of the references are. It should be borne in mind that THIS is the current state of the art, not what was going on between the wars or in the 19th century and is now old hat for philosophers as well as knowledge engineers and cognitive scientists of all kinds.

The distinction between hypothesis, theory and fact are not only outdated, but have been found to be useless in practice. AI IS used today, and it follows Minsky and Popper.


Putnam's view on the philosophy of science - a few fascinating snippets.

In the book "Men of Ideas" (BBC 1978 ISBN 0 563 17436 6), Bryan Magee interviews fifteen of the top philosophers of the time (not including Popper, though he was alive then, and indeed kicking).

Hilary Putnam (a bloke, at Harvard) was chosen for the discussion on the philosophy of science. (Curiously, they managed 13 pages of dialogue without once mentioning Popper, though he featured heavily in Magee's summaries elsewhere in the book.)

Coming in on the second page of the discussion:

MAGEE: [...][F]or two or three hundred years educated Western man thought of the universe and everything in it as consisting of matter in motion - this was all there was [...]. 'Science' consisted it finding out more about this matter, and its structure, and the laws of its motion, by a special method known as 'scientific method'. If it went on doing this for long enough it would eventually find out everything there is to find out. This view of science has now been abandoned by scientists - but that fact seems not yet to have got through to the majority of non-scientists.

PUTNAM: I think it started to break down with Einstein. If I may bring in the history of philosophy, Kant did something which is very relevant to our discussion. He questioned the corresppondence theory of truth. Before Kant, no philosopher doubted that truth was correspondence to reality, or 'agreement' with reality. The image was of knowledge as a mirror, or copy. But Kant said: 'It isn't so simple. There is the contribution of the thinking mind.' Of course, knowledge isn't made up by the mind - Kant wasn't an idealist. It isn't all a fiction. But it isn't just a copy either. What we call 'truth' depends on both what there is (the way things are) and on the contribution of the thinker. [...] I think Einstein came to a similar view - that there is a human contribution, a conceptual contribution, to what we call 'truth'. Scientific theories are not simply dictated to us by the facts.

MAGEE: I think some people may find that idea puzling. How can it be that what is true and what is not true can not depend on what the facts are, regardless of the human mind?

PUTNAM: Well, let me use an analogy with vision. We tend to think tht what we see just depends on what's out there; but the more one studies vision, either as a scientist or as a painter, the more one discovers that what's called 'vision' involves an enormous amount of interpretation. The colour we see as red is not the same colour, in terms of wavelength, at different times of the day, so that even in what we think of as our simplest transaction with the world - just looking at it - we are interpreting.

MAGEE: And in fact we bring to bear on the world a whole conceptual and categorical scheme that we're not usually aware of unless we consciously turn our gaze inward and start examining it.
[...]
What you're saying, then, is that the categories in terms of which we see the world and interpret our experience - and the frameworks within which we organise our observations [JJ: Note implicit reference to Minsky's ('75) 'Frame' concept, a type of compound concept...], which are what we usually like to call 'the facts' - are contributed by us, and this means that the world as conceived by science is partly constituted by facts outside ourselves but also partly constituted by categories which we bring to it.

PUTNAM: Yes. [He gives the example of the wave/particle duality, and contiunues...] Philosophers have started to speak of 'equivalent descriptions' [...].

[They discuss the ideas that science steadily progresses by one theory giving way to another indefinitely, and that certainty is something claimed now only by the very religious: by scientists its very possibility is denied. Putnam's reply to this is "It may be" - but his uncertainty here is due to his not wanting to generalise about the religious.]

[...]

MAGEE: [...] [N]ow that we've so completely altered our conception of science, must it not also be the case that the difference between science and non-science can no longer be accepted as being what it used to be thought to be?

PUTNAM: I think that is true and culturally very important. The harm that the old picture of science does is this: if there is a realm of absolute fact that scientists are gradually accumulating, then everything else appears as non-knowledge, as something to which 'true' and 'false' can't properly apply. [...]

MAGEE: [...] But if this is so, [...] what is the point of continuing to use the word 'science'? Does it still demarcate anything which there are valid intellectual reasons for seeking to demarcate?

PUTNAM: I don't think it does. Distinguishing science from non-science made a lot of sense, given the old view that there is such a thing as 'the scientific method', and that what makes something a science is that it uses it rather consciously and rather deliberately, while in a non-science it is either used unconsciously (as in learning to cook) or not used at all. In fact however, it does not seem there is such a thing as the scientific method. There are of course general maxims for empirical enquiry - for example, the very fact that we can speak of empirical enquiry reflects one of them: 'Don't try to figure out in a purely a priori way how nature works'. That natural science is a posteriori, not a priori, was Bacon's contribution to modern thought, and it was a great one. And I mentioned two of the corollaries of this a minute ago - 'test your ideas', and 'remember that your ideas are corrigible'. But which theories we should test, and which we should regard as 'too crazy' even to test; when a theory has been sufficiently tested to warrant provisional acceptance, and when it has been tested enough to be relied on, at least until a better theory comes along; these are all matters which in practice scientists decide partly on the basis of tradition (imitating what Kuhn calls 'paradigms', ie previous examples of successful practice in their fields) and partly on the basis of intuition. It does not seem that there is any mechanical rule that factors out human psychology, human intuitive judgements of 'reasonableness' and 'plausibility', such that science could in principle be done by a computing machine which just followed that rule, given enough time and enough data. But 'science' was traditionally supposed to be different from ordinary practical knowledge just by virtue of the fact that there was a method, 'inductive logic', which 'science' consciously followed. To say both that there is a sharp line between science and non-science, and that that the method which is supposed to draw this line is fuzzy and, indeed, incapable of any but the most vague and general description at the present time seems silly to me.

By the way, attempts to formalize induction are very much a failure [JJ: we'll see ;-)] - 'inductive logic', if there is such a thing, has not been programmed successfully on a computer. The development of deductive logic in the last hundred years, and the development of the computer, have brought home very dramatically just what a different position we are in with respect to the proof in the mathematical sciences, which we can state rigorous canons for, and proof in the inductive sciences, where all we can state are the sort of maxims I alluded to before.

MAGEE: As you say, this means tht the traditional idea of there being any particular scientific method at all has to be revised. For a long time people were clear about what the one and only scientific method was. You carried out closely-controlled and carefully-measured observations, and when you had thus gathered a great quantity of reliable data you proceed by inductive logic to formulate a general theory which would be explanatory with respect to the observed phenomena; you then thought up a crucial experiment to test your theory; and if the theory passed the test it was verified. For something like two centuries or more this method, and this method alone, was thought to be 'scientific'. But now that our whole conception of science has altered, it is not so much that some different method has come to be regarded as 'the' method, but rather that there is no longer thought of as being only one single valid approach to all scientific problems.

[...]

So you think 'science' is now simply a term for the pursuit of knowledge?

PUTNAM: That's right.

MAGEE: [...] [C]an we now turn to consider what you and your colleagues, the philosophers of science, do?

PUTNAM: Well, part of what we do [...] is fairly technical investigation of specific scientific theories. We look at [quantum mechanics, relativity, Darwinian evolution] very closely. [He mentions that some philosophers of science have different views regarding induction.] [A]nd there are philosophers like myself who have an in-between position, who think that there is something in the notion of 'scientific method', that there are clear examples, but that there is more or less a continuum between scientific knowledge and ordinary unformalized knowledge that we don't dignify with the honorific name of 'science', and that we mustn't think of scientific method as a mechanical rule, an algorithm, that one can apply to get scientific knowledge. I would say that these issues - the nature of truth, the nature of scientific method, and also whether there is any such thing as necessary truth in science (any conceptual contribution which is eternal, and not subject to revision) - are central and live issues in philosophy of science today.

MAGEE: Who are you doing all this for primarily - scientists, or philosohers? I ask because I've taken part in attempts to bring scientists and philosophers together to discuss these issues. The attempts have usually failed, and for the same reason: the scientists weren't sufficiently interested. But whereas the bulk of working scientists, it seems to me, don't really care much about these matters, it's conspicuous that great scientists tend to be among the exceptions. Many of the path breakers who actually made the scientific revolution of this century have written books of philosophical reflection about it - Einstein, Max Born, Niels Bohr, Heisenberg and (my particular favourite in this genre) Shrödinger, just to mention a few. Nevertheless, as I say, the majority of working scientists don't seem to be very interested.

PUTNAM: Well, I'd first of all say that we are writing for the philosophically-interested layman, for the reader of philosophy. I don't view philosophers of science as giving direct advice to scientists, just as I think moral philosophers are ill-advised [...to give moral advice]. On the other hand, I do think that scientists tend to know the philosophy of science of fifty years ago, and perhaps this isn't a bad thing; perhaps this time lag, this cultural lag, has some value in weeding out what they shouldn't be paying attention to. Of course, it is annoying to a philosopher to encounter a scientist who is sure that he needn't listen to any philosophy of science and who then produces verbatim ideas which you can recognise as having been popular in 1928.

[...]

MAGEE: A further point [...] is that although computers were originally constructed by conscious analogy with the human mind, as they become more sophisticated we begin to learn things about the human mind from them. So on the one hand our construction of computers, and on the other hand what they tell us about ourselves, develop by a process interactive growth. It is worth noting, too, that what we have here is an interaction not just between philosophy and science but between philosophy and technology.

PUTNUM: I agree. This is one area, by the way, in which philosophers are in close contact with scientists. The fields of linguistics, cognitive psychology, computer science, and philosophy of language exhibit a constant and healthy interaction.

[...]




Back to main web page